A few-shot learning model trains global class prototypes on both base and novel classes, with a registration module and sample synthesis, evaluated on Omniglot and miniImageNet.
An empirical study and analysis of generalized zero- shot learning for object recognition in the wild
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2019 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Few-Shot Learning with Global Class Representations
A few-shot learning model trains global class prototypes on both base and novel classes, with a registration module and sample synthesis, evaluated on Omniglot and miniImageNet.